I do not know if the article author/cartographer ever studied linguistics or phonetics, but this is the main takeaway from the map for me, a linguist, the pattern is the message, not the somewhat imprecise data points.
As was common for Chinese romanization systems from before the 20th century, it used 't' for unaspirated /t/ (which we would use use 'd' for today in systems like Pinyin for Mandarin or Peng'im for Teochew), and used 'th' for aspirated /tʰ/ (which we sould use 't' for today in systems like Pinyin or Peng'im).
This isn't an archaic miss transliteration, it's just an alternative transliteration strategy. In many languages that primarily use the latin alphabet, the phonemes associated with 't' and 'd' are /t/ and /d/, primarily distinguished by voice instead of aspiration (where aspiration is allophonic), so it's logical that the creators of earlier romanization systems focused on preserving that voice distinction, even if it's less common today for a variety of reasons.
So failed look at archaic miss transliterations, kind of reminiscent of the entire theme of the article. The example map being built on an argument from a bad assumption (a historical error as a pretext for argument).
Perfect irony.